Ccn2a-FGFR1-SHH signaling is necessary for intervertebral disc homeostasis and regeneration in adult zebrafish

Intervertebral discs (IVDs), fibrocartilaginous tissue between adjacent vertebrae, are present in all vertebrates. In mammals, including humans, IVDs consist of centrally placed large vacuolated notochordal cell (NC) populated nucleus pulposus (NP), which is encapsulated by annulus fibrosus (AF), a multilayered angularly arranged lamellar collagenous structure consisting of fibroblasts (Hashizume, 1980). Furthermore, AF is subdivided into two zones; outer annulus fibrosus (OAF), made up of small, tightly packed cells and inner annulus fibrosus (IAF), made up of elongated cells (Rufai et al., 1995).

IVD degeneration (IVDD) is considered a major reason for back, neck and appendage pain, putting a considerable socio-economic burden on the clinical system (Cheung et al., 2009). Historically, disc degeneration was believed to be associated only with aging (Buckwalter, 1995); however, recent studies suggest that it is also linked to genetic (Eskola et al., 2010) and health-related (Teraguchi et al., 2017) backgrounds. Physical analysis of human discs has shown that IVDD starts from the NP region (Pearce et al., 1987). However, the molecular and cellular mechanisms involved in IVD homeostasis and degeneration are poorly understood. Irrespective of the reason, alterations in extracellular matrix (ECM) composition and cell morphology are visible in degenerating discs (Roughley, 2004). Thus, understanding the role and working cascades of ECM molecules in IVD maintenance and regeneration will help develop therapeutic strategies to counter IVDD.

Cellular communication network factor 2 (CCN2), an ECM molecule, plays a role in the progression of fibrotic diseases (Jaffa et al., 2008; Mori et al., 1999; Tamatani et al., 1998) as well as in tissue regeneration (Mokalled et al., 2016; Mukherjee et al., 2021; Riley et al., 2015) in a context- and tissue microenvironment-dependent manner. In mice, notochord-specific deletion of Ccn2 leads to an early onset of IVDD (Bedore et al., 2013). An in vitro study showed that the combinatorial application of TGFβ1 and CCN2 promotes rat tail disc-derived NP cell survival (Matta et al., 2017). In contrast, Wu et al. suggest that TGFβ-induced CCN2 secretion leads to notochordal cell death in vitro (Wu et al., 2019). Thus, the cellular and molecular functions of CCN2 in IVDD remain to be explored.

In this study, we sought to explore the role of Ccn2a, a zebrafish orthologue of human CCN2, in IVD homeostasis and regeneration in zebrafish. Our histology revealed that cellular morphology and arrangement in zebrafish IVDs are similar to that in mammalian counterparts. We find that ccn2a is expressed in adult IVDs. Although ccn2a mutants are viable adults, ccn2a^−/− IVDs show diminished cell proliferation and survival at 4 months post-fertilization (mpf), leading to an early onset of IVDD. Interestingly, in aged degenerated IVDs, ectopic expression of Ccn2a promotes OAF cell proliferation and NP cell survival. Through genetic and biochemical approaches, our findings reveal that Ccn2a maintains the cell turnover in adult IVDs and can induce regeneration in degenerated IVDs by augmenting FGFR1-SHH signaling. This study enriches our understanding of the cellular and molecular mechanisms through which Ccn2a maintains IVD homeostasis and its potential therapeutic implications to promote regeneration in degenerated human discs.

Although a handful of data are available about chordacentra morphogenesis (Haga et al., 2009; Lleras Forero et al., 2018; Wopat et al., 2018), the cellular arrangement in adult zebrafish IVDs remains to be determined. Our histological analysis identified that, at 12 days post fertilization (dpf) (∼6 mm length), notochordal sheath cells (NSCs) from the junctional area of adjacent chordacentra (future IVD forming region) protrude into the notochordal cell zone (Fig. S1A,B). At 20 dpf (∼8 mm length), sagittal sections through the intervertebral region showed three distinct types of cellular arrangements: an outermost tightly packed single layer of small cells, followed by one or two layers of elongated cells arranged in parallel with the spine and innermost large cells (Fig. S1C,C′). The number of layers of small cells in the outermost compact zone and elongated cells in the adjacent zone increase with the progression of age towards adulthood (Fig. S1D,D′). Based on the similarity of cellular morphology and arrangements between adult mammals and zebrafish, we named the outermost area covered by three or four layers of compact cells as the OAF, the adjacent area filled with four or five layers of elongated cells as the IAF, and the innermost region filled with large cells as the NP (Fig. S1D,D′). IVDs and the tissue connecting IVDs are covered with a monolayer of epithelial-like cells, presumably the NSCs (Fig. S1C-D′). Overall, these observations suggest that the cellular arrangement of zebrafish IVDs is majorly aligned with mammalian IVDs.

In situ hybridization on zebrafish embryos reported that both paralogs of CCN2: ccn2a and ccn2b are expressed in notochord (Fernando et al., 2010). However, the spatiotemporal expression pattern of ccn2a and ccn2b in notochordal tissue and IVDs remained unknown. Semi qPCR results reveal that ccn2a and ccn2b expression were first detected at 9 h post-fertilization (hpf) and 12 hpf, respectively (Fig. 1A). Whole-mount in situ hybridization showed that both ccn2a and ccn2b transcripts are expressed in the notochord at 3 dpf (Fig. 1B). Interestingly, transverse sections of the 3 dpf whole-mount in situ hybridized embryos showed that ccn2a transcripts were localized in the NSCs, whereas ccn2b transcripts were detected in the NCs (Fig. 1B). In situ hybridization on the sagittal sections of 3 mpf zebrafish showed that both ccn2a and ccn2b were prominently expressed in the OAF and NSCs (Fig. 1C). Weak expression was visible in the IAF and NP cells (Fig. 1C). Altogether, in situ hybridization data suggest that ccn2a and ccn2b are heterogeneously expressed in the adult IVDs.

Next, the TgBAC(ccn2a:EGFP)^ari1 reporter line (Mukherjee et al., 2021) was employed to identify the ccn2a-expressing cell populations. BACccn2a:EGFP expression mimicked endogenous ccn2a mRNA expression. Transverse sections of the 3 dpf embryo identified BACccn2a:EGFP expression in the NSCs (Fig. 1D). In line with the in situ analysis, in 3 mpf zebrafish, prominent BACccn2a:EGFP expression was seen in the OAF cells and NSCs (Fig. 1E). Weak and scattered EGFP expression was detected in IAF and NP cells (Fig. 1E). Furthermore, immunolocalization studies detected ubiquitous CCN2 protein localization throughout the adult IVD (Fig. 1F and Fig. S2A). Furthermore, although 24 mpf old BACccn2a:EGFP zebrafish IVD showed similar spatial EGFP expression to 3 mpf young adults, decreased EGFP expression was observed in 24 mpf IVD compared with their 3 mpf young siblings (Fig. S2B). In addition, the qPCR analysis showed decreased ccn2a transcripts in the vertebral tissues of 24 mpf animals compared with 3 mpf (Fig. S2C). Overall, ccn2a and ccn2b transcripts are expressed in the NSCs and NCs of the embryonic notochord, respectively; in adult IVD, both paralogs are predominantly expressed in the OAF and NSCs, CCN2 protein is localized ubiquitously, and Ccn2a expression is decreased in aged IVDs.

Preceding results showed that ccn2a and ccn2b transcripts are expressed in the developing notochord and adult IVDs. We, therefore, sought to explore their role in IVD morphogenesis and homeostasis. To study this, we employed a loss-of-function ccn2a mutant allele (the ctgfa^bns50 allele hereafter ccn2a⁻) (Mokalled et al., 2016) and generated an eight-nucleotide deletion ccn2b mutant allele (ccn2b^ari2 allele hereafter ccn2b⁻) (Fig. S3A-C). qPCR analysis showed a ∼50% decreased ccn2b transcripts in 30 hpf ccn2b^−/− relative to the wild-type embryos, suggesting that ccn2b⁻ mRNA is unstable (Fig. S3D). Both ccn2a^−/− and ccn2b^−/− are adult viable. Although ccn2b^−/− morphologically remained indistinguishable from wild-type siblings, at least until 12 mpf (Fig. S3E), ∼60% of ccn2a^−/− animals showed a curved-body phenotype between 10 and 12 mpf (Fig. 2A), which prompted us to examine the spine structure of 12 mpf zebrafish. Alcian Blue and Alizarin Red S (AB/AR) staining showed the presence of intervertebral gaps, the space occupied by the IVDs, in 12 mpf wild-type as well as ccn2b^−/− animals (Fig. S3F); however, visible intervertebral gaps were not detected in all analyzed ccn2a^−/− siblings (Fig. 2B), suggesting that the IVD thinning phenotype in ccn2a^−/− zebrafish is 100% penetrant.

Furthermore, to find the onset of the IVD thinning phenotype, we explored the spine morphology from 1- to 6-mpf by AB/AR staining. ccn2a^−/− showed the presence of intervertebral gaps similar to wild-type siblings until 4 mpf (Fig. 2C). At 6 mpf, despite ccn2a^−/− being morphologically similar to wild types (Fig. S3G), a curved spine lacking visible intervertebral gaps in all investigated ccn2a^−/− (Fig. 2C) indicated a possibility of IVDD. Furthermore, µCT imaging confirmed reduced intervertebral gaps in 6 mpf ccn2a^−/− compared with their wild-type siblings (Fig. 2D). Oegema et al. suggested that increased fibronectin expression and the fragmented appearance of the fibronectin protein in the IVD are the hallmarks of mammalian IVDD (Oegema et al., 2000). Therefore, we explored the possibility of IVDD in ccn2a^−/− by analyzing the immunolocalization of fibronectin (FN1) and transcript levels of fibronectin1 paralogs (fn1a and fn1b) in 6 mpf IVDs. Immunostaining on sagittal sections of wild-type zebrafish revealed that fibronectin is localized in a lamellar pattern in the IAF, with a scattered expression in the rest of the IVD (Fig. 2E). In contrast, noticeable fragmented fibronectin was detected in the IAF of age-matched ccn2a^−/− IVDs (Fig. 2E). Furthermore, fluorescence intensity analysis revealed a ∼6-fold increased fibronectin expression in ccn2a^−/− relative to wild type (Fig. 2E,F). Similarly, qPCR analysis showed increased fn1a expression in vertebral tissues of ccn2a^−/− compared with wild-type siblings, while fn1b expression remained unaltered (Fig. 2G). Thus, these results suggest that deficiency of Ccn2a leads to IVD degeneration in zebrafish.

Prospective reasons for IVDD in ccn2a^−/− could be inefficient cell proliferation and/or increased cell death. As cellular homeostasis in wild-type zebrafish IVDs is yet to be explored, we examined cell proliferation and cell death in 4 mpf adult wild-type zebrafish IVDs (Fig. S4A). Interestingly, EdU incorporation data showed proliferating cells only in the OAF region (Fig. 3A). In contrast, TUNEL⁺ cells were observed in the NP region (Fig. 3B). Moreover, EdU⁺ or TUNEL⁺ cells were undetected in the IAF cells and NSCs (Fig. 3A,B). Taken together, our study suggests that only OAF cells can proliferate whereas NP cells are prone to die in 4 mpf adult zebrafish IVDs.

Next, we analyzed and compared cell proliferation and cell death indices in IVDs of 4 mpf animals when intervertebral gaps were visible in ccn2a^−/−, similar to wild type (Fig. 2C). An EdU incorporation assay on sagittal vertebral tissue sections identified a ∼50% reduction in OAF cell proliferation in ccn2a^−/− compared with wild-type siblings (Fig. 3C,D). TUNEL assay on sagittal vertebral tissue sections revealed a ∼40% increased NP cell death in ccn2a^−/− compared with wild-type siblings (Fig. 3E,F). Furthermore, we asked whether the cell turnover in ccn2a^−/− IVDs is affected during morphogenesis. No significant difference was visible in cell proliferation or in cell death (Fig. S4B-F) between wild-type and ccn2a^−/− siblings at 1 mpf, indicating normal IVD morphogenesis in ccn2a^−/−. Furthermore, as in 4 mpf adults, in 1 mpf juvenile animals, BrdU⁺ proliferating cells and TUNEL⁺ cells were found only in the OAF and NP regions, respectively (Fig. S4C and E). Altogether, these observations suggest that although ccn2a^−/− appears to develop normal IVDs, Ccn2a is required for OAF cell proliferation and for NP cell survival in adult zebrafish IVDs.

In mammals, several growth factors and collagens are associated with IVDD. Increased TGFβ1 (Peng et al., 2009) and BMP2 (Hollenberg et al., 2021) expression were observed in degenerating human IVDs. It has also been reported that Shh positively regulates NP cell proliferation in embryonic and postnatal mouse IVDs (Choi and Harfe, 2011; Dahia et al., 2012), and its expression decreases with age progression (Winkler et al., 2014). Besides the growth factors, collagens are also associated with IVD maintenance in mammals (Trefilova et al., 2021). As adult ccn2a^−/− showed disc degeneration, we intended to analyze expression levels of growth factor-and collagen-coding genes in 4 mpf zebrafish vertebral tissues. qPCR analysis revealed a reduction in transcript levels of shha (∼36%) and shhb (∼44%), whereas those of other analyzed growth factor- and collagen-coding genes remained indistinguishable (Fig. 3G) in vertebral tissues of ccn2a^−/− compared with wild type.

Next, in situ hybridization analysis detected ubiquitous shha and shhb expression in the wild-type IVDs (Fig. 3H). Notably, consistent with the qPCR data, reduced expression of shha and shhb was detected in the ccn2a^−/− IVDs without altering the spatial expression pattern (Fig. 3H). Moreover, immunohistochemistry found ubiquitous localization of Shha in wild-type IVDs with a ∼60% decreased expression in ccn2a^−/− IVDs compared with wild type (Fig. S5A,B). Decreased expression of shha and shhb transcripts, as well as Shha protein prompted us to explore whether SHH signaling is affected in ccn2a^−/−. qPCR analysis showed decreased expression of SHH target genes gli1, ptch1 and ptch2 (Bonifas et al., 2001; Chuang et al., 2003; Litingtung et al., 1998) in vertebral tissues of ccn2a^−/− compared with wild type (Fig. 3I). Overall, Ccn2a positively regulates SHH signaling by modulating shha and shhb expression in adult zebrafish IVDs.

Next, we intended to explore the underlying signaling cascade responsible for Ccn2a-mediated regulation of shha and shhb expression in adult IVDs. Studies suggest that Shh is a downstream target of Fgfr1 and Fgfr2 during limb (Verheyden et al., 2005) and heart (Lavine, 2006) morphogenesis. Dahia et al. have shown that pan-FGFR signaling is active in the AF region but not in the NP region of the postnatal and adult mouse IVDs (Dahia et al., 2009). Moreover, FGFR1 (Li et al., 2008; Yan et al., 2015) and FGFR3 (Yan et al., 2015) are expressed differentially in degenerated human discs. Our preceding results showed that IVDD is initiated at 4 mpf in ccn2a^−/−. Therefore, a comparative qPCR was performed to identify whether FGF receptor coding genes are differentially expressed in 4 mpf ccn2a^−/− vertebral tissues. Our assessment revealed an ∼2-fold increased expression of fgfr1b, a zebrafish paralog of human FGFR1, in vertebral tissues of ccn2a^−/− compared with wild type, while expression of fgfr1a, another paralog of human FGFR1, and other FGFRs remained unaffected (Fig. 4A). Furthermore, in situ hybridization showed ubiquitous fgfr1a and fgfr1b expression in wild-type IVDs (Fig. 4B). Consistent with qPCR data, fgfr1b transcripts were predominantly induced in the ccn2a^−/− IVDs compared with wild type, while fgfr1a expression remained unaffected (Fig. 4B). Altered fgfr1b expression in ccn2a^−/− IVDs, indicates that Fgfr1b signaling could be involved with the IVD degeneration. Furthermore, published evidence and altered expression of shha, shhb and fgfr1b led us to hypothesize that FGFR1 signaling is associated with the shha and shhb expression in adult zebrafish IVDs. To investigate this, we used pharmacological and genetic approaches. For the pharmacological approach, we inhibited the FGFR1 signaling for 4 days in 4 mpf wild-type zebrafish by intraperitoneal injection of PD166866, a FGFR1 specific inhibitor (Fig. 4C) (Panek et al., 1998). To identify whether PD166866 inhibits FGFR1 signaling in the zebrafish vertebral tissues, we explored the expression of zebrafish orthologs of FGFR1 signaling target genes [CCND1, CCND2A, CCND2B (Koziczak et al., 2004), ETV4 and ETV5A (DeSalvo et al., 2021)] upon PD166866 treatment. qPCR analysis of the vertebral tissues showed decreased ccnd1, ccnd2a, ccnd2b, etv4 and etv5a gene expression in the PD166866-treated animals compared with the control (Fig. S6A), indicating PD166866 inhibits FGFR1 signaling in zebrafish vertebral tissues. Furthermore, qPCR analysis showed a ∼40% decreased expression of shha and shhb (Fig. 4D), whereas transcript levels of fgfr1b and ccn2a remained unaffected upon PD166866 treatment (Fig. 4E), indicating that FGFR1 signaling induces shha and shhb expression, but not ccn2a, in adult IVDs.

To further confirm, we examined the transcript levels of shha and shhb in vertebral tissues of ccn2a^−/− upon ectopic activation of FGFR1 signaling. We carried out FGFR1 gain-of-function experiments using Tg(hsp70:ca-fgfr1) zebrafish, which ubiquitously expresses a constitutively active form of Xenopus Fgfr1 upon heat shock (Marques et al., 2008). We raised ccn2a^−/− on hsp70:ca-fgfr1 background (hereafter ccn2a^{−/−,fgfr1−ca}). Wild-type, ccn2a^−/− and ccn2a^{−/−,fgfr1-ca} siblings were subjected to heat shock from 2.5 to 4 mpf, and RNA was extracted from vertebral tissues 24 h after the final heat shock (Fig. 4F). Consistent with the pharmacological inhibition data, qPCR analysis revealed an ∼2-fold increase in expression of FGFR1 downstream target genes (ccnd1, ccnd2a, ccnd2b, etv4 and etv5a) (Fig. S6B) and an ∼2.5-fold increase in shha and shhb expression upon ectopic expression of the constitutively active form of Xenopus Fgfr1 in ccn2a mutants, while fgfr1b expression remained unaltered (Fig. 4G). Thus, pharmacological inhibition and gain-of-function studies revealed that FGFR1 signaling induces shha and shhb expression in vertebral tissues. Overall, these data suggest that shha and shhb are common downstream targets of Ccn2a and FGFR1 signaling. Notably, in ccn2a^−/− IVDs, despite the increased fgfr1b expression (Fig. 4A,B), decreased FGFR1 target gene (shha and shhb) expression (Fig. 3G,H) was observed, which could be rescued by ectopic expression of the active form of Fgfr1 (Fig. 4G) indicating that Ccn2a is required for FGFR1 activation in adult IVDs.

As ccn2a^−/− IVDs showed altered cellular homeostasis, and Ccn2a, as well as FGFR1 signaling, induce shha and shhb expression, we analyzed the cell proliferation and cell survival in adult IVDs upon inhibition of FGFR1 signaling. EdU incorporation and TUNEL assay revealed that 4 days of PD166866 treatment (Fig. 4H) resulted in decreased OAF cell proliferation (Fig. 4I,J) and increased NP cell death (Fig. 4K,L), similar to the phenotype observed in the adult ccn2a^−/− IVDs. Taken together, these observations indicate that, similar to Ccn2a, FGFR1 signaling induces shha and shhb expression, OAF cell proliferation and NP cell survival in adult IVDs.

As our data showed that Ccn2a, as well as FGFR1 signaling, induce OAF cell proliferation and NP cell survival, and that Ccn2a is not a downstream target of FGFR1 signaling, we were curious to investigate whether ectopic expression of a constitutively active form of Xenopus Fgfr1 can rescue the cellular phenotype of ccn2a^−/− IVDs. To address this, wild-type, ccn2a^−/−, and ccn2a^{−/−,fgfr1−ca} siblings were subjected to heat shock from 2.5 to 4 mpf, followed by EdU incorporation and TUNEL assay (Fig. 5A). Analysis revealed a ∼95% increase in OAF cell proliferation (Fig. 5B,C) and a ∼75% decrease in NP cell death (Fig. 5D,E) in ccn2a^{−/−,fgfr1-ca} compared with ccn2a^−/−. Notably, OAF cell proliferation and NP cell death in ccn2a^{−/−,fgfr1-ca} were comparable with the wild-type IVDs (Fig. 5B-E). Moreover, extended ectopic activation of FGFR1 signaling from 2.5 to 6 mpf could maintain the intervertebral gaps in ccn2a^−/− (Fig. 5F,G), suggesting that ectopic activation of FGFR1 signaling can restore IVDs by augmenting OAF cell proliferation and NP cell survival in ccn2a^−/−.

Next, we explored whether Ccn2a or FGFR1 signaling-mediated OAF cell proliferation and NP cell survival are regulated through the SHH pathway. To test this, cyclopamine, a SHH signaling inhibitor (Incardona et al., 1998) was injected to 4 mpf wild-type zebrafish for 4 days (Fig. S7A), followed by gene expression, cell death and cell proliferation analysis were performed. qPCR analysis of vertebral tissues revealed decreased expression of SHH target genes gli1, ptch1 and ptch2 without any effect on shha and shhb transcripts in cyclopamine-injected compared with DMSO-injected siblings, indicating SHH signaling inhibition (Fig. S7B). Moreover, SHH signaling inhibition leads to suppression of OAF cell proliferation (Fig. 5H,I) and increased NP cell death (Fig. 5J,K), indicating the positive role of SHH signaling in OAF cell proliferation and NP cell survival in adult IVDs. Overall, these results indicate that, presumably, Ccn2a and FGFR1 signaling induce OAF cell proliferation and NP cell survival through the SHH pathway in adult IVDs.

Micro-CT analysis showed signs of disc degeneration in 12 mpf wild-type zebrafish (Kague et al., 2021). Hence, we compared the shha and shhb expression, OAF cell proliferation, and NP cell death in 4 and 12 mpf wild-type vertebral tissues. We found decreased shha and shhb expression (Fig. 6A), reduced OAF cell proliferation (Fig. 6B,C) and increased NP cell death (Fig. 6D,E) in 12 mpf compared with 4 mpf wild-type IVDs. Further, similar to juvenile and 4 mpf adults, in 12 mpf animals, EdU⁺ proliferating cells and TUNEL⁺ cells were found only in the OAF and NP regions, respectively (Fig. 6B,D). Overall, the observed cellular phenotype and genetic regulation in 12 mpf IVDs were similar to 4 mpf ccn2a^−/− IVDs, and decreased ccn2a expression was observed in aged IVDs (Fig. S2B,C). Hence, we sought to assess the potency of Ccn2a to promote regeneration in aged IVDs. We used hsp70:ctgfa-FL-2a-EGFP (henceforth ccn2a^{+/+,hsp70:ctgfa}) transgenic zebrafish (Mokalled et al., 2016), which ubiquitously express Ccn2a upon heat shock. Heat shock treatments for 7 days (Fig. 6F) resulted in an ∼4-fold increase in ccn2a expression along with increased shha and shhb expression in vertebral tissues of 12 mpf ccn2a^{+/+,hsp70:ctgfa} relative to ccn2a^+/+ siblings (Fig. 6G). Moreover, EdU incorporation and TUNEL assay (Fig. 6F) showed that ectopic Ccn2a expression resulted in induction of OAF cell proliferation (Fig. 6H,I) and NP cell survival (Fig. 6J,K), suggesting that in aged vertebral tissues, Ccn2a induces shha and shhb expression, and can promote cellular processes involved in IVD regeneration.

Our study identified that Ccn2a and FGFR1 signaling induce shha and shhb expression, NP cell survival and OAF cell proliferation in adult IVDs, and that Ccn2a is not a downstream target of FGFR1 signaling. Thus, we intended to explore whether Ccn2a exerts its regulation through the FGFR1 signaling cascade in adult IVDs. We combined genetic and pharmacological approaches on 12 mpf animals to dissect this. Control groups (ccn2a^+/+ and ccn2a^{+/+,hsp70:ctgfa}) received DMSO, while a test group (ccn2a^{+/+,hsp70:ctgfa}) received PD166866 injection from day 4 to day 6, all groups received heat shock from day 0 to 6, and animals were sacrificed on day 7 (Fig. 6L). Animals used for the cell proliferation assay received a single EdU injection on day 6 (Fig. 6L). qPCR analysis revealed an increased expression of shha and shhb (Fig. 6M) as well as of the FGFR1 downstream target genes ccnd1, ccnd2a, ccnd2b, etv4 and etv5a (Fig. S8A) in the vertebral tissues upon overexpression of Ccn2a. Furthermore, PD166866 treatment suppressed the ectopic Ccn2a-mediated induction of expression of shha and shhb (Fig. 6M) and FGFR1 signaling target gene (Fig. S8B) without affecting ccn2a expression (Fig. 6M), indicating that Ccn2a induces shha and shhb expression through FGFR1 signaling.

Next, to confirm the link between Ccn2a and FGFR1 signaling, we used SU5402, another FGFR1 signaling inhibitor (Mohammadi et al., 1997) (Fig. S8C). Gene expression analysis showed decreased expression of FGFR1 target genes ccnd1, ccnd2a, ccnd2b, etv4 and etv5a (Fig. S8D), indicating SU5402 inhibits FGFR1 signaling in zebrafish vertebral tissues. Moreover, in the SU5402-treated animals, decreased shha and shhb expression was observed, while fgfr1b and ccn2a remained unaffected (Fig. S8E). Next, we performed the pharmacological inhibition in combination with the genetic overexpression of Ccn2a (Fig. S8F). In line with PD166866 inhibition, SU5402 treatment suppressed the ectopic Ccn2a-mediated increased FGFR1 downstream target genes (Fig. S8G), as well as shha and shhb expression without affecting the ccn2a expression (Fig. S8H). Overall, data confirm that Ccn2a induces shha and shhb expression through FGFR1 signaling.

Furthermore, in Ccn2a overexpression animals, ectopic Ccn2a mediated increased OAF cell proliferation and NP cell survival was suppressed upon PD166866 treatment (Fig. 6N-Q). As short-term ectopic Ccn2a induced OAF cell proliferation and NP cell survival in aged IVDs, we intended to explore the effect of ectopic Ccn2a expression on degenerated IVD over an extended period. Thus, 12 mpf old no-background (control) and hsp70:ctgfa animals were given daily 1 h heat shock for 90 days (Fig. 6R). Skeletal staining of the animals after 90 days of treatment showed visibly increased intervertebral gaps in Ccn2a-overexpressing animals compared with their no-background siblings (Fig. 6S). Altogether, in degenerated IVDs, Ccn2a induces regeneration by inducing shha and shhb expression, OAF cell proliferation and NP cell survival through FGFR1 signaling.

Preceding results revealed that Ccn2a plays a central role in the maintenance and regeneration of IVDs through FGFR1-SHH signaling in adult zebrafish. Hence, we explored the molecular crosstalk between Ccn2a and the zebrafish orthologs of FGFR1 (Fgfr1a and Fgfr1b) using a proximity ligation assay (PLA). In vitro synthesized C-terminal Myc-tagged full-length ccn2a (Ccn2a-FL^Myc) mRNA was co-injected with C-terminal FLAG-tagged- fgfr1a (Fgfr1a^FLAG) or fgfr1b (Fgfr1b^FLAG) mRNA into a one-cell stage zebrafish egg (Fig. 7A). Immunostaining against Myc and FLAG on injected embryos confirmed tagged Ccn2a-FL, Fgfr1a and Fgfr1b expression (Fig. S9A,B). PLA on 30 hpf embryos showed that zebrafish Ccn2a-FL interacts with Fgfr1a and with Fgfr1b (Fig. 7B,C).

Ccn2a is a modular protein with four interaction domains and a hinge region at which it cleaves into a pro-fibrotic N-terminal fragment (NTF) and a proliferative C-terminal fragment (CTF) (Fig. S9C) (Grotendorst and Duncan, 2005; Robinson et al., 2012); our PLA showed the interaction between C-terminal tagged Ccn2a-FL and Fgfr1a or Fgfr1b (Fig. 7B,C), indicating that the CTF of Ccn2a interacts with Fgfr1a or Fgfr1b. To further explore whether the NTF of Ccn2a interacts with Fgfr1a and Fgfr1b, in vitro synthesized mRNA of C-terminal Myc-tagged IGFBP and vWC domain coding NTF of ccn2a (Ccn2a-NTF^Myc) was co-injected with Fgfr1a^FLAG or Fgfr1b^FLAG mRNA (Fig. 7A). Immunostaining against Myc and FLAG on injected embryos confirmed tagged Ccn2a-NTF, Fgfr1a and Fgfr1b expression (Fig. S9D,E). PLA analysis on 30 hpf embryos indicated that the NTF of Ccn2a does not interact with Fgfr1a (Fig. S9F) or Fgfr1b (Fig. S9G). Overall, C-terminal tagged full-length Ccn2a, but not the C-terminal tagged NTF of Ccn2a, interacts with Fgfr1a and Fgfr1b, indicating that Ccn2a interacts with the ligand-receptor complex of Fgfr1a, as well as Fgfr1b, through its C-terminal fragment.

This study provides previously unreported insights into the mechanism by which Ccn2a maintains the tissue integrity of IVDs and its potential role in regeneration in degenerated aged IVD. ccn2a is expressed by the OAF cells and NSCs, and Ccn2 protein is localized ubiquitously in adult IVDs. Mechanistically, Ccn2a promotes OAF cell proliferation and NP cell survival by inducing FGFR1-SHH signaling in young as well as aged adult IVDs. Based on the above findings, we propose that Ccn2a is an essential component of IVD homeostasis and can induce IVD regeneration.

In agreement with notochord-specific Ccn2 knockout mice (Bedore et al., 2013) data, we found that global loss of Ccn2a in zebrafish leads to an early IVDD. It is interesting to note that Ccn2 global knockout mice die after birth (Ivkovic et al., 2003); however, ccn2a^−/− zebrafish survive adults. Unlike mammals, zebrafish have two paralogs of CCN2 – ccn2a and ccn2b – both of which are expressed in larval notochord. qPCR analysis identified an ∼4-fold increase in expression of ccn2b in ccn2a^−/− larvae compared with wild type (Fig. S10), indicating that, possibly in the absence of Ccn2a, increased Ccn2b contributes to the normal development of the ccn2a^−/−.

Our cell proliferation and cell death analysis in the juveniles, 4 mpf young adults, as well as in 12 mpf aged animals found that zebrafish IVDs carry distinct cell proliferation and cell death regions. Like mammals (Dahia et al., 2009), the NP cells of adult zebrafish IVDs are prone to die; however, unlike mammals, they are not proliferating. In zebrafish embryos, upon injury, dead notochordal cells become replenished by the cells derived from NSCs (Garcia et al., 2017). Because, in adult IVDs, only the OAF cells are proliferating (Fig. 3A), presumably a subpopulation of OAF cells migrates, differentiates and replenishes different cell populations in IVDs. Furthermore, studies in rabbits and humans have shown the presence of stem cells in the cartilaginous endplates that proliferate and differentiate into different cell types of IVDs (Henriksson et al., 2009). Thus, it will be interesting to know whether, like mammals, zebrafish IVDs have an external source of cells to replenish the dead cells or whether the proliferating OAF cells have multipotent characteristics, and also whether these are the only sources of all types of cells in zebrafish IVDs.

In adult IVDs, FGFR1 signaling induces shha and shhb expression (Fig. 4G). Surprisingly, in ccn2a^−/− IVDs, despite increased fgfr1b transcripts (Fig. 4A,B), its downstream targets are downregulated (Fig. 3G,H). Thus, it is possible that, as Ccn2a induces FGFR1 signaling, in ccn2a^−/−, decreased levels of Shha and Shhb induce fgfr1b expression via a feedback loop mechanism. However, changes in the shha and shhb transcripts levels upon pharmacological inhibition of FGFR1 signaling in wild-type animals (Fig. 4D) or ectopic expression of a constitutively active form of Fgfr1 in ccn2a^−/− (Fig. 4G) did not affect the transcript levels of fgfr1b (Fig. 4E,G), ruling out the presence of feedback loop between the fgfr1b and its downstream molecules in adult vertebral tissues. Furthermore, in vivo PLA showed that Ccn2a binds to Fgfr1a as well as to Fgfr1b, suggesting Ccn2a can signal through both the paralogs and is an integral part of the FGFR1 ligand-receptor complex for the receptor activation. Notably, altered expression of fgfr1b in ccn2a^−/− indicates that, presumably, Ccn2a signals through the Fgfr1b as its preferred receptor. The absence of a feedback loop between the Fgfr1b and its downstream targets and upregulation of fgfr1b in the ccn2a^−/− IVDs indicates that, plausibly, a Ccn2a-dependent active form of the Fgfr1b ligand-receptor complex regulates fgfr1b expression through an unknown mechanism that needs to be elucidated.

It is believed that in IVDs, CCN2 is a pro-fibrotic molecule because CCN2 is upregulated in degenerated human discs (Peng et al., 2009). In contrast, our study on zebrafish found that overexpression of Ccn2a induces regenerative cellular processes in degenerated IVDs. PLA showed that only CTF, which is known to be a pro-proliferative fragment (Grotendorst and Duncan, 2005; Robinson et al., 2012), interacts with Fgfr1a as well as Fgfr1b, supporting the hypothesis that Ccn2a is a pro-regenerative rather than a pro-fibrotic molecule in IVDs. It is possible that, in aged human IVDs, induction of CCN2 is a regenerative response; however, the amount of CCN2 is insufficient to induce effective regeneration in degenerated discs. In future, it will be interesting to explore whether an external supply of a higher amount of CCN2 can induce regeneration in degenerated mammalian IVDs. Furthermore, Ccn2a promotes IVD regeneration by inducing FGFR1 signaling in zebrafish IVDs (Fig. 6). Thus, comparative studies exploring the effect of FGFR1 agonists or CCN2 on mammalian IVD regeneration may open new avenues to develop therapeutic strategies to manage IVD regeneration in humans.

In conclusion, our study shows that Ccn2a is a crucial secreted molecule in IVD maintenance. Moreover, induction of cell proliferation and suppression of cell death by ectopic Ccn2a in degenerated IVDs shows its potential to induce IVD regeneration. This study increases our understanding of the cellular homeostasis of the zebrafish IVDs and of the essential role of Ccn2a-FGFR1-SHH signaling in IVD maintenance and regeneration, which suggests that CCN2 could be used therapeutically to promote regeneration in degenerated human discs.

Zebrafish care procedures and protocols used in this study were as per guidelines of the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), Government of India, and were approved by The Institute Animal Ethics Committee (IAEC).

Zebrafish were maintained in a state-of-the-art zebrafish aquarium as previously described (Aleström et al., 2020). Animals ranging from fertilized eggs to 24 mpf were used in this study. The genotype of animals used is specified in the figures and figure legends. Clutch mate adult male fish of a similar size were used for the experiments associated with PD166866 or cyclopamine treatment. For the rest of the experiments, clutch mate control and test fish of a similar size and equal sex distribution were used for each age group.

TALEN was used to generate the ccn2b^ari2 mutant allele. TALEN constructs for left and right arms were designed to target exon 2, with a 17-19 bp long repeat-variable di-residue (RVD) binding sequence and a 15 bp long spacer; left arm sequence, TGATTGCCCAGACGAGA; right arm sequence, TAGGCACCAGTCTAGTGTT. Left and right arm TALEN pair constructs were generated with the help of the Golden Gate TALEN assembly strategy (Cermak et al., 2015). TALEN mRNAs were synthesized using an in vitro mMESSAGE mMACHINE kit, following the manufacturer's instructions. 200 pg mRNA of each arm was injected into one-cell stage eggs to generate mutants. Injected eggs were raised and screened for germline transmission. Indel detection at the target locus was performed by High-Resolution Melting Curve Analysis (HRMA) of the gDNA isolated from caudal fins of adult F1 animals. Indels were analyzed by sequencing the PCR amplified product using a primer pair (forward, CTGCTCTCAGCCCTGTGATT; reverse, CAGCAGCTACAGCCGTCTAA) binding to the flanking region of the TALEN target. The ari2 allele (referred to as ccn2b⁻), harboring an 8 nucleotide deletion, was selected and used in this study. For genotyping and gene expression analysis by qPCR, the following primer pair was used: forward, TGATTGCCCAGACGAGAGCCC; reverse, CAGCAGCTACAGCCGTCTAA. qPCR was performed to check ccn2b⁻ transcript stability. Transcript levels of ccn2b⁻ in ccn2b^−/− were quantitated relative to the average expression level ccn2b⁺ in wild-type controls and were normalized to ef1α expression levels.

According to the manufacturer's instructions, total RNA from embryos and adult vertebral tissues was extracted using TRIzol reagent (Invitrogen). Total RNA was isolated from 30 pooled embryos/biological replicates for embryonic stages. For adult vertebral tissues, animals were sacrificed and vertebral tissues were isolated manually by removing the head, organs, fins, skin, skeletal muscles and spinal cord as much as possible without disturbing the vertebral column, and processed for total RNA isolation. Furthermore, 1-4 µg of total RNA was reverse transcribed into cDNA using MMLV reverse transcriptase (Invitrogen, 28025013). The prepared cDNA was used to perform qPCR in a PCR Max Real-time PCR detection system (Cole Parmer) using iTaq Universal SYBR Green Supermix (Bio-Rad, 1725124) and gene-specific primer pair (Table S1). Transcript levels of genes analyzed in test samples were quantitated relative to the control sample average expression level and normalized to ef1a. All reactions were run in triplicate. Gene expression fold changes were calculated using the ΔCt method. For adults, each sample represents one animal.

For making riboprobes against zebrafish ccn2b, fgfr1a, fgfr1b, shha and shhb mRNA, partial cDNA was amplified from embryonic cDNA using gene-specific primers (Table S2) and cloned into the pGEM-T Easy vector (Promega) to create pGEMTeasy-Zccn2b, pGEMTeasy-Zfgfr1a, pGEMTeasy-Zfgfr1b, pGEMTeasy-Zshha and pGEMTeasy-Zshhb constructs.

For the proximity ligation assay (PLA), coding sequences of full-length zebrafish ccn2a, a N-terminal fragment of ccn2a, full-length fgfr1a and full-length fgfr1b were amplified from embryonic cDNA using gene-specific primers (Table S3). Full-length ccn2a and a N-terminal fragment of ccn2a (amino acids 1-162) were cloned into the pCMV-Myc-C vector (Clontech), while full-length fgfr1a and fgfr1b were cloned into the p3xFLAG-CMV-14 vector (Merck). Full-length ccn2a, a N-terminal fragment of ccn2a, full-length fgfr1a and full-length fgfr1b along with C-terminal MYC or FLAG tag were amplified from respective constructs using specific primers and subcloned into the pCS2+ vector to create FL-Zccn2a-MYC, NTF-Zccn2a-MYC, FL-Zfgfr1a-FLAG and FL-Zfgfr1b-FLAG (Table S3), respectively, for the synthesis of in vitro capped mRNA.

Digoxigenin (DIG) labeled riboprobes against ccn2a, ccn2b, fgfr1a, fgfr1b, shha and shhb were synthesized using linearized pGEMTeasy-Zccn2b, pGEMTeasy-Zfgfr1a, pGEMTeasy-Zfgfr1b, pGEMTeasy-Zshha and pGEMTeasy-Zshhb constructs, respectively, and T7 RNA polymerase (Roche, 10881767001).

For whole-mount in situ hybridization, from around 24 hpf, embryos were treated with 0.2 mM 1-phenyl-2-thiourea to prevent pigmentation, and embryos were fixed in 4% paraformaldehyde, washed twice with 0.2% Tween-20 in 1×PBS, dehydrated and processed for whole-mount in situ hybridization using digoxigenin-labeled riboprobes against ccn2a (Mukherjee et al., 2021) and ccn2b. For sectioning after signal development, embryos were washed with deionized water, fixed in 4% paraformaldehyde, dehydrated in 30% sucrose and embedded in the tissue freezing medium (Leica, 14020108926). 10 µm transverse sections through the trunk were cut (Leica, CM1520), air dried, washed in PBS, mounted in the Kaiser's gelatin glycerol (Merck) and imaged with an Olympus BX53 bright-field microscope.

For in situ hybridization on sections, fixed tissues were processed for wax sectioning and in situ hybridization was performed as described previously (Lepilina et al., 2006). Sagittal sections (12 µm) through the trunk were mounted on a charged glass slide (Thermo Fisher Scientific), rehydrated, permeabilized with proteinase K at 37°C, acid washed and hybridized with DIG-labelled riboprobes (1 µg/ml in hybridization buffer) at 65°C overnight. On the second day, sections were briefly washed with wash buffer, blocked in blocking buffer and incubated overnight with an alkaline phosphatase-conjugated anti-DIG antibody (Roche, 11093274910) at 4°C. On the third day, sections were washed and the signal was detected with BM Purple staining solution (Roche, 11442074001). Furthermore, sections were dehydrated through ethanol grades, xylol and xylene, and mounted in Entellan (Merck) for imaging.

Skeletal staining was performed as previously described (Kessel and Gruss, 1991). Briefly, fish were sacrificed at desired time points, deskinned and fixed in 96% ethanol for 24 h and subsequently in acetone for 48 h. Fixed samples were incubated in the Alcian Blue 8GX (Merck) and Alizarin Red S (Merck) staining solution for 6 h and 2 h, respectively, at 37°C. Stained samples were placed under tap water to wash off the excess stain and cleared in 1% KOH overnight at room temperature. The following day, samples were cleared in grades of glycerol (50% and 70%) in 1% KOH and then 100% glycerol. Lateral views of the zebrafish spines of the cleared samples were imaged with a stereoscope (Leica M205FA) or a digital single-lens reflex camera (Canon 600D).

Six mpf wild-type and ccn2a^−/− fish were anesthetized in 0.02% tricaine in the system water. The caudal part of the wild-type and ccn2a^−/− vertebral column were scanned using a SkyScan 1276 micro-CT scanner (Bruker) with voxels sized at 10 µm. Detailed geometric analysis was performed using an X-ray source of 200 kV, 60 µA with an Al 0.5 mm filter. Images were reconstructed using NRecon Reconstruction Software (Micro Photonics).

For heat-shock experiments, Tg(hsp70l:Xla. Fgfr1,cryaa:DsRed)^pd3 raised in ccn2a^−/− background, and Tg(hsp70:ctgfa-FL-2a-EGFP) raised in a wild-type background were used. These transgenic animals and their respective controls received heat shock by incubation in a preheated water system at 37°C for 1 h (maximum 6 fish per liter water) at 24 h intervals for the desired period. The age and genotype of animals are mentioned in each figure. Animals were sacrificed 24 h after the last heat shock and processed for qPCR, histological analysis and skeletal staining.

At desired time points, fish were sacrificed, cut into two halves at the anal fin and the caudal part of each fish was fixed in 4% PFA overnight at room temperature. Fixed caudal parts were washed with PBS several times, deskinned and decalcified in 0.5 M EDTA (pH 8.0) in PBS for 4 days at room temperature. For wax sectioning, tissues were washed with PBS and dehydrated in a series of ethanol grades (50%, 70%, 90% and 100%) before embedding in wax, and 10 µm tissue sections were cut with a microtome (Bright Instruments, Leica Histocore Autocut). For cryosectioning, fixed tissues were washed with PBS, dehydrated in 30% sucrose in PBS, embedded in tissue-freezing medium (Leica, 14020108926) and 10 µm tissue sections were cut with a cryomicrotome (Leica, CM1520).

For EdU incorporation analysis, fish were anesthetized with 0.02% tricaine in system water and 10 µl of 10 mM EdU was injected intraperitoneally. Fish were sacrificed 24 h post-EdU injection, fixed in 4% PFA overnight at RT, decalcified in 0.5 M EDTA (pH 8.0) in PBS for 4 days at room temperature and cryosectioned before EdU labeling. EdU labeling was performed on the 10 µm sagittal tissue sections through the vertebral column using Click-iT EdU Cell Proliferation Kit (Thermo Fisher Scientific, C10340), according to the manufacturer's instructions, and DAPI (1 µg/ml in PBS) was used to stain nuclei. For imaging, sections were mounted in Mowiol 4-88 Reagent (Merck).

For BrdU incorporation analysis, 1 mpf animals were treated with BrdU (100 µg/ml) in system water for 2 days before fixing in 4% PFA overnight at room temperature. Fixed animals were processed for cryosectioning. For BrdU detection, tissue sections were air-dried and washed in PBS, refixed in 4% PFA and treated with 1 N HCl in PBS at 37°C for 1 h. Later, sections were blocked using blocking solution [5% goat serum (MP Biomedicals)/0.5% Triton X-100/PBS] for 1 h and incubated with BrdU antibody (1:200; Abcam, Ab6326) overnight at 4°C. Primary immune complexes were detected using goat, anti-rat Alexa Flour 488-conjugated secondary antibodies (1:400; Thermo Fisher Scientific, A11006), and DAPI (1 µg/ml PBS) was used to detect nuclei. For imaging, stained sections were mounted in Mowiol 4-88 Reagent (Merck, 475904). Six animals from test and control groups, from two independent experiments, were used to quantify OAF cell proliferation indices. At least two IVDs from each animal and four sections from each IVD were analyzed for quantification.

A fluorescence terminal deoxynucleotidyl transferase dUTP Nick-End Labeling (TUNEL) assay was performed to detect dead cells in the IVDs. For this fluorescence TUNEL assay, 10 µm sagittal cryosections of the caudal part of the desired genotype and aged zebrafish were prepared as described under ‘Tissue sectioning’. The TUNEL assay was performed on the cryosections using a Click-iT TUNEL Alexa Fluor 647 Imaging kit (Thermo Fisher Scientific, C10247) following the manufacturer's instructions. DAPI (1 µg/ml PBS) was used to detect nuclei. For imaging, sections were mounted in Mowiol 4-88 Reagent (Merck) and images were acquired with a Leica SP8 confocal microscope. Six animals from test and control groups from two independent experiments were used to quantify NP cell death indices. At least two IVDs from each animal and four sections from each IVD were analyzed for quantification.

Immunolocalization studies were performed on the tissue sections, as described previously (Patra et al., 2017). Briefly, PBS-washed tissue cryosections were refixed, permeabilized with 0.5% Triton X-100, blocked for 1 h in a blocking solution [5% goat serum (MP Biomedicals)/0.5% Triton X-100/PBS], and incubated with primary antibodies against CTGF (1:100; Abcam, Ab6992) or Shha (1:100; Anaspec, S-55574) or fibronectin (1:600; Merck- F-0635) overnight at 4°C. The goat, anti-rabbit Alexa Flour 647-conjugated secondary antibody (1:400; Thermo Fisher Scientific, A32733) detected primary immune complexes. DAPI (1 µg/ml PBS) was used to detect nuclei. Sections were mounted in Mowiol 4-88 Reagent (Merck) before imaging with a Leica SP8 laser scanning microscope.

Wheat germ agglutinin (WGA) staining labeled the cell membrane. For WGA labeling, cryosections through the vertebral tissues were incubated in Alexa Flour 633-conjugated WGA (2 µg/ml in PBS) (Biotium, 29024) for 1 h at room temperature before permeabilization. Sections were stained with DAPI to label nuclei before embedding in the mounting medium (Mowiol 4-88, Merck). Images were captured on a Leica SP8 laser scanning microscope, and images were processed and analyzed with a LAS X (Leica). At least two IVDs from each animal and four sections from each IVD were analyzed for quantification.

Fish were anesthetized in 0.02% tricaine in system water for body weight determination and intraperitoneal injection. For body weight determination, anesthetized fish were kept on tissue paper to remove water from the body surface and weigh the fish. The drug dose for individual fish was calculated based on the body weight. The next day, intraperitoneal injections of the drugs were made on anesthetized fish immobilized into a wet foam holder. PD166866 (Merck, PZ0114) was dissolved in DMSO at a stock concentration of 25 mM. It was diluted in PBS and injected intraperitoneally at 370 µg/g body weight. Cyclopamine (Merck, C4116) was dissolved in DMSO at a stock concentration of 10 mM. The stock was diluted in PBS and injected intraperitoneally at 41.3 µg/g body weight. The injection volume of the chemical or DMSO (vehicle control) was adjusted to 10 µl with 1×PBS and injected intraperitoneally once a day for 4 days. SU5402 (Merck, SML0443) was dissolved in DMSO at a stock concentration of 17 mM. The inhibitor was added to the system water at a volume to make the working concentration of 17 µM; the same volume of DMSO was added as vehicle control. Four animals were maintained in 100 ml system water with DMSO or SU5402. Animals received fresh treatment every 24 h for 3 days. Control and treated animals were maintained in separate tanks during the experiment. Animals were sacrificed 24 h after the last injection and processed for qPCR or histological analysis.

Stained tissue sections were imaged with a Leica SP8 confocal laser scanning microscope, keeping all the parameters fixed for scanning the test and control samples. The acquired confocal z-stacks were processed and analyzed with LAS X (Leica) or GIMP (GIMP Development Team) software.

Proximity ligation assay (PLA) was performed to identify the interaction between zebrafish Ccn2a and Fgfr1a or Fgfr1b. C-terminal Myc-tagged full-length ccn2a (Ccn2a-FL^Myc), C-terminal Myc-tagged N-terminal fragment of ccn2a (Ccn2a-NTF^Myc), C-terminal FLAG-tagged full-length fgfr1a (Fgfr1a^FLAG) and C-terminal FLAG-tagged full-length fgfr1b (Fgfr1b^FLAG) mRNAs were synthesized using linearized FL-Zccn2a-MYC, NTF-Zccn2a-MYC, FL-Zfgfr1a-FLAG and FL-Zfgfr1b-FLAG, respectively, and mMESSAGE mMACHINE SP6 Transcription Kit (Thermo Fisher Scientific, AM1340) according to the manufacturer's instructions. In vitro synthesized mRNAs of Ccn2a-FL^Myc or Ccn2a-NTF^Myc were co-injected with Fgfr1a^FLAG or Fgfr1b^FLAG mRNA in one-cell stage embryos. Embryos were harvested at 30 hpf, fixed in 4% PFA in PBS overnight at 4°C. Around 50% of the fixed embryos were processed for immunostaining, and the remaining embryos were processed for the proximity ligation assay (PLA).

Whole-mount immunostaining was performed on 30 hpf fixed embryos to detect the expression of injected tagged mRNAs. Fixed embryos were washed thrice in PBS, permeabilized with 0.5% Triton X-100 in PBS and blocked for 1 h in blocking solution [5% goat serum (MP Biomedicals) and 0.5% Triton X-100 in PBS]. Subsequently, embryos were incubated with primary antibodies [mouse anti-Myc (1:250; Merck, M4439) and rabbit anti-FLAG (1:250; Cell Signaling Technology, 14793S)]. Primary immune complexes were detected using goat anti-mouse Alexa Flour 488- (Thermo Fisher Scientific, A21042) and goat anti-rabbit Alexa Flour 647- (Thermo Fisher Scientific, A32733) conjugated secondary antibodies (1:400). DAPI (1 µg/ml in PBS) was used to detect nuclei. For imaging, embryos were flat-mounted in Mowiol 4-88 Reagent (Merck) and imaged with a Leica SP8 confocal laser scanning microscope.

After confirmation of the expression of the injected mRNAs, PLA was performed on the rest of the injected clutch mates using Duolink In Situ Orange Starter Kit Mouse/Rabbit (Merck, DUO92102) according to the manufacturer's instruction. In brief, fixed embryos were processed and treated with the primary antibodies similar to the above-mentioned whole-mount immunostaining. The next day, embryos were washed in buffer A twice for 5 min each at room temperature and incubated with the PLA probe solution for 1 h at 37°C. Subsequently, embryos were washed in buffer A twice for 5 min each at room temperature and incubated in the ligation solution for 30 min at 37°C. Furthermore, embryos were washed in buffer A twice for 5 min each at room temperature and incubated in amplification solution for 100 min at 37°C. After washing in buffer B twice for 10 min each at room temperature, embryos were incubated in DAPI (1 µg/ml PBS, Merck) to detect nuclei. Embryos were flat-mounted on a glass slide in Duolink In Situ Mounting medium, optical sections were captured with a Leica SP8 confocal laser scanning microscope and images were processed with the LAS X (Leica) or GIMP (GIMP Development Team) software.

An unpaired Student's t-test was performed to analyze statistical differences in gene expression, cell proliferation indices and cell death indices. Statistical significance was considered at P<0.05 for all analyses. Data were processed with GraphPad Prism7 software. Values are represented as mean±s.e.m.

We are grateful to Kenneth D. Poss for providing the Tg(hsp70:ca-fgfr1) transgenic zebrafish and to Mahendra Sonawane for critical reading of the manuscript. We thank Satish Bojja for the excellent fish care and the Agharkar Research Institute for internal support. We thank Mr M. B. Daware and Mrs R. J. Londhe for their support in laser scanning microscopy. We thank the Director, Dr Dhakephalkar, MACS-Agharkar Research Institute, Pune, India for providing infrastructure and core facilities.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201036.reviewer-comments.pdf